search

monarch-initiative / monarch-disease-ontology-RETIRED

THIS IS THE OLD REPO: Use this one instead: https://github.com/monarch-initiative/mondo-build
https://github.com/monarch-initiative/mondo-build
17 stars 9 forks source link

NCIT_C3200 | Lung Neoplasm mapping #365

Closed nicolevasilevsky closed 6 years ago

nicolevasilevsky commented 7 years ago

Not sure if this mapping is correct, but I think it may be okay. Depends on how cancer should be mapped to NCIT.

NCIT_C3200 Lung Neoplasm SuperClassOf DOID_1324 lung cancer 0.7999715434

image

cmungall commented 7 years ago

Yes, I agree. As this was given a high probability and your evaluation agrees we don't need to make a ticket for all of these.

However, since we have this open, it's worth looking at the bottom left there. Note the low probability placement of the MESH term. I would have made this equivalent. I'm trying to figure why this wasn't done...

cmungall commented 7 years ago

so mesh place nscLC as a subtype of bronchogenic:

   is_a MESH:D002283 ! Carcinoma, Bronchogenic
      is_a MESH:D002289 ! Carcinoma, Non-Small-Cell Lung ***

in NCIT they are direct sibs:

image

I am generally inclined to trust NCIT over MESH. I suspect MESH is doing it this was as many nscLCs do arise in peripheral tissue, but yet they are wrong to classify it this way as there are exceptions?

In DO, they are more distant sibs:

image

This highlights an issue with strict location based classification: the bronchus is typically not considered part of the lung. bronchiogenic carcinoma arises in the epithelium of bronchus/iole and migrates to the lung